{"id":"https://openalex.org/W7166841334","doi":"https://doi.org/10.18653/v1/2026.acl-long.1640","title":"LBLLM: Lightweight Binarization of Large Language Models via Three-Stage Distillation","display_name":"LBLLM: Lightweight Binarization of Large Language Models via Three-Stage Distillation","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166841334","doi":"https://doi.org/10.18653/v1/2026.acl-long.1640"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.1640","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1640","pdf_url":"https://aclanthology.org/2026.acl-long.1640.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.1640.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5080001189","display_name":"Siqing Song","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Siqing Song","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139751318","display_name":"Chuang Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chuang Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134296409","display_name":"Yong Lang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yong Lang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139715533","display_name":"Yi Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yi Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139827396","display_name":"Xu-Yao Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu-Yao Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.83112565,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"35470","last_page":"35484"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.24860000610351562,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.24860000610351562,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.2117999941110611,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11550","display_name":"Text and Document Classification Technologies","score":0.07100000232458115,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/distillation","display_name":"Distillation","score":0.3273000121116638},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.3111000061035156},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.26499998569488525},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.2533000111579895}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6410999894142151},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5440000295639038},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.35510000586509705},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.3273000121116638},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.319599986076355},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3111000061035156},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2793000042438507},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.26499998569488525},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.2533000111579895},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.24729999899864197}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.1640","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1640","pdf_url":"https://aclanthology.org/2026.acl-long.1640.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.1640","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1640","pdf_url":"https://aclanthology.org/2026.acl-long.1640.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320321133","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166841334.pdf","grobid_xml":"https://content.openalex.org/works/W7166841334.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Deploying":[0],"large":[1],"language":[2,115,120],"models":[3],"(LLMs)":[4],"in":[5,60,148],"resource-constrained":[6],"environments":[7],"is":[8],"hindered":[9],"by":[10],"heavy":[11],"computational":[12],"and":[13,51,63,83,90,119,134],"memory":[14],"requirements.We":[15],"present":[16],"LBLLM,":[17],"a":[18,28,39,100,153],"lightweight":[19],"binarization":[20,106],"framework":[21,33],"that":[22,124],"achieves":[23],"effective":[24,136],"W(1+1)A4":[25],"quantization":[26,31,52,68,110,127],"through":[27,54],"novel":[29],"threestage":[30],"strategy.The":[32],"proceeds":[34],"as":[35],"follows:":[36],"(1)":[37],"initialize":[38],"high-quality":[40],"quantized":[41],"model":[42],"via":[43],"PTQ;":[44],"(2)":[45],"quantize":[46,72],"binarized":[47],"weights,":[48],"group-wise":[49],"bitmaps,":[50],"parameters":[53],"layer-wise":[55],"distillation":[56],"while":[57],"keeping":[58],"activations":[59,73],"full":[61],"precision;":[62],"(3)":[64],"training":[65,88],"learnable":[66],"activation":[67,84],"factors":[69],"to":[70,74],"dynamically":[71],"4":[75],"bits.This":[76],"decoupled":[77],"design":[78],"mitigates":[79],"interference":[80],"between":[81],"weight":[82],"quantization,":[85],"yielding":[86],"greater":[87],"stability":[89],"better":[91],"inference":[92],"accuracy.LBLLM,":[93],"trained":[94],"only":[95],"using":[96],"0.016B":[97],"tokens":[98],"with":[99],"single":[101],"GPU,":[102],"surpasses":[103],"existing":[104],"state-of-the-art":[105],"methods":[107],"on":[108,160],"W2A4":[109],"settings":[111],"across":[112],"tasks":[113],"of":[114,128],"modeling,":[116],"commonsense":[117],"QA,":[118],"understanding.These":[121],"results":[122],"demonstrate":[123],"extreme":[125],"low-bit":[126],"LLMs":[129],"can":[130],"be":[131],"both":[132],"practical":[133],"highly":[135],"without":[137],"introducing":[138],"any":[139],"extra":[140],"high-precision":[141],"channels":[142],"nor":[143],"rotational":[144],"matrices":[145],"commonly":[146],"used":[147],"recent":[149],"PTQ-based":[150],"works,":[151],"offering":[152],"promising":[154],"path":[155],"toward":[156],"efficient":[157],"LLM":[158],"deployment":[159],"resource-limited":[161],"situations.":[162]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}